Performance-Based Dowel DesignLift-truck design changes require a new look at joint durability

Serviceability is the name of the game with floor slabs that will have lift truck traffic. The most vulnerable places on such a floor slab are the joints. The joints break down when a lift truck moves toward the joint, deflecting down the edge of the slab panel it is on, then bumping against the joint face of the adjacent, slightly higher, panel.

Relying on aggregate interlock for long-term load transfer at the contraction joints of such slabs is impractical, as we have previously noted (see Ref. 1 and 2). The American Concrete Institute (ACI) publications have been recommending dowels at joints for a number of years. ACI 360R-06 “Design of Slabs on Ground” (see a summary of this new document) states that “Doweled joints are recommended when positive load transfer is required,” and ACI 302.1R-96 and ACI 302.1R-04 “Guide for Concrete Floor and Slab Construction” have similar recommendations.

Most slab thickness design procedures assume that load is transferred between adjacent slab panels. Our experience is that to protect the joints proper load transfer is especially important when significant lift truck traffic is anticipated. Thus, doweled contraction joints should be used to minimize joint spalling due to lift truck traffic, minimize lift truck maintenance cost, and share the load to prevent the higher stresses resulting from the loading of free edges. But when dowels are used, the slab designer should consider the properties of the dowel system specified, which include its geometry, installation tolerances, and bond-breaking material, along with the cost of the dowel system. If only one of these properties are compromised, then severe and costly problems could occur.

This article is a continuation of an article we wrote in 1998 (see Ref. 3), where we discussed the many benefits of plate dowels. Tapered plate dowels have been in use for over four years on a number of projects. In this article, we will discuss the benefits of using tapered plate dowels in contraction joints and provide design recommendations for the size and spacing of these dowels for industrial floors to accommodate lift truck loadings. These design recommendations are based on both strength and serviceability criteria for lift truck loadings and are more rational than the traditional method of selecting the dowel size and spacing based on slab thickness.

Most of the significant dowel research and corresponding recommendations (such as in References 4 and 5) were done in the 1940s and 1950s. These recommendations were for round dowels and for highway traffic loadings with wheels spaced 5 to 9 feet apart. The dowel recommendations in ACI 302.1R-04 are based on these highway types of loads and may not be conservative enough for some lift truck loadings, while being too conservative for some other types of loads. For industrial floor slabs where lift trucks will be used, the wheel loads can be higher than on a highway—the tires are a hard solid material (as opposed to the large, soft pneumatic tires used for highway traffic), the load contact area is over a smaller area (due to the hard solid tire material), and the wheels are at a much closer spacing (18 to 42 inches).

One of the earliest uses of tapered plate dowels was this pavement section at the Atlanta Bonded Warehouse distribution center in Kennesaw, Ga. After four years of constant truck traffic there is no damage at the contraction joints even thought the slab is only 6 inches thick

The recommendations for round dowels for highway traffic loadings were developed with the objective of limiting the bearing pressure of the dowel on the concrete. But there are other dowel design requirements that are important for industrial floors slabs with lift trucks, such as the relative deflection between the slab panels, the effect of curling on the deflection of slab panels with dowels, and how curling affects the distribution of the force in the dowels due to the wheel loads. None of these were considered.

We have developed extensive computer programs, along with using a commercially available program, to analyze the forces in the dowels and to determine the relative differential deflection between the slab panels. The model used is shown in Fig. 1.

We used a nonlinear analysis using a finite plate element with a compression-only spring for the base support to simulate the curled-slab profile, which will lose base contact near the joints. This condition is common for slabs on ground (as noted in Reference 6) and will affect the magnitude and distribution of the forces in the dowels. Depending on the magnitude of the wheel load, the curled slab may or may not come back into contact with the base; this condition is accounted for in the computer model.

As part of this analysis, we have made the following assumptions:

1. Concrete strength. The compressive strength of the concrete is 3500 psi. This is the strength recommended by ACI 302.1R-04 for steel-troweled floor slabs and hard-wheeled traffic.

2. Subgrade. The modulus of sub-grade reaction for the base and soil support system is 150 pounds per cubic inch. This is a typical value for short-term loadings such as from lift trucks. Fortunately, the analysis for the dowel forces and relative deflection between the slabs is relatively insensitive to large changes in this value so it need only be approximate.

3. Dowel support properties. There has been much discussion (such as in Fig. 1—Basis of the computer model used to design tapered plate dowels. Refs. 4 and 5) and some direct testing (see Ref. 7) to establish the concrete modulus of dowel support. The direct testing indicated that “a single value of the modulus of dowel support could not be used to back-predict the experimentally observed dowel deformations along the length of the dowel” but “overall joint load-displacement behavior appeared to be linear” (from Ref. 7). Testing (Ref. 5) also indicated that the joint load-displacement was linear after the initial looseness was taken up by the initial loading and a condition of full bearing was established.

The concrete modulus of dowel support also seems to vary with the width of the dowel (Ref. 4 and 5). Fortunately, the concrete modulus of dowel support value is relatively insensitive to the analysis and need only be approximate; we have chosen a value of 1,500,000 pounds per cubic inch. This value is what was used in Reference 4 for all dowel sizes, including the wider dowels, and is a little less than the value determined by testing for the wider dowel in Reference 5. The testing did indicate that the concrete modulus of dowel support varied some with the concrete compressive strength, but the value we have selected is representative of the 3500 psi concrete recommended for industrial floor slabs.

Because the dowel is tapered, we determined the dowel properties beyond the saw-cut location tolerance of 2 inches (see Fig. 2) on the smaller side. We found that the properties varied only slightly when compared with the properties using the average width of the dowel, and that this variation was in the same range as the other design variables. Alternating the directions of the dowels, as shown in Fig. 3, also helps minimize this small difference. Therefore, we used the average dowel width in developing our recommendations.

Table 1: Vertical dowel spring values

We used the material properties mentioned above to determine the vertical spring in the computer model that represents the dowel stiffness that is used to transmit the wheel shear load to the adjacent panel and to determine the deflection and stresses in the dowel. These spring values (see Table 1), along with the deflection and stresses in the dowel, were determined using the equations in our previous article (Ref. 3).

4. Slab curling. We have used an equivalent shrinkage gradient of 45° F between the top and bottom of the slab to establish the curling profile of the slab. This value was chosen based on the many slab profiles that we have taken for 6-inch-thick slabs with 15-foot joint spacing where the corner of the slab panel would be approximately 1/8 inch to ¼ inch higher than the center of the panel. This value is somewhat higher than the 30° F gradient that we used in some of our previous analyses, which were based on much earlier data, and is probably an indication that concrete shrinkage has increased somewhat over the years (described in detail in Refs. 6 and 8).

5. Loads from lift trucks. We have used two load cases for each of the different lift trucks. For the first load case, the lift truck was positioned on top of the dowel, and for the second load case, it was positioned between the dowels. The force in the dowel for the load case that produced the maximum differential deflection between the slab panels was used as the maximum allowable load for the dowel. Typically, for dowels at close spacing, the lift truck position on top of the dowel produced the maximum force in the dowel and the maximum deflection. For dowels spaced farther apart, the lift truck positioned between the dowels produce the maximum deflection. Even though the force in the dowel was less with the lift truck positioned between the dowels, the deflection of the slab spanning between the dowels became significant. Therefore, the allowable loads for the dowels spaced farther apart were reduced to account for this transverse slab deflection and to meet our maximum differential deflection criteria.

We used typical lift truck load data for two of the most common types of lift trucks with solid tires: the traditional (counterbalanced) lift truck and the pallet lift truck. Our experience is that only about 75% of the rated load capacity of the lift truck is moved with a regular frequency and rarely does the lift truck move the full rated capacity. Because the design criterion is based on fatigue, it would be more rational to base the selection of the dowels on the most common repetitive loading. Therefore, we have used 75% of the lift truck's rated capacity for our design recommendations. For the few facilities where the full rated capacity of the lift truck is moved on a frequent basis, the data in Table 2 can be used to show the ratio the values in the design graphs.

6. Joint width. We assumed a maximum joint opening size of 0.20 inches, which should be sufficient for normal joint spacings used with typical concrete mixes.

7. Slab thickness. Three common slab thicknesses were used for the analysis: 6, 8, and 10 inches, with joint spacings of 15, 18, and 21 feet, respectively.

8. Dowel spacing. Five different dowel spacings were used for the analysis: 12, 18, 24, 30, and 36 inches.

9. Load-dowel combinations. To simplify the number of possible combinations with multiple plate dowel sizes and the different load cases, a conservative assumption was made to use the stiffer spring value of the ¾-inch plate for all of the load cases. This conservative assumption increased the force in the dowels by 15% for the worst case, but in most cases, only by 3% to 8%.